In a server/client model, a client accesses a server via communication network and inputs a command including a predetermined query and the server responds to the user terminal with a response corresponding to the query.
However, the user may inputs the command for another purpose instead of receiving the response.
For example, a search server providing a search service may statistically analyze a search query of a user and use the search query in generating search result list in order to provide a search result list according to the preference of the user.
Since a search result “frequently” selected by a user receiving a predetermined search result list is considered to have high relation with a search query and be highly preferred by users, the search server adds the preference to the search result such that the search result is preferentially provided to a user with respect to the search query.
In this case, “search result selection” of the user is used as “a query for receiving information associated with the search result or accessing a web page associated with the search result”. On the other hand, “search result selection” is used as “a standard for generating a search result list”. A user knowing the mechanism in the search server may repeatedly select the same search result from the search result list provided in response to a predetermined search query such that the preference is added to the search result selected by the user. Of course, in case that the user repeatedly selects a search result with malicious intent and the preference is added to the search result, an original purpose to preferentially provide a search result having high relation with the search result and preferred by users can not be achieved.
Accordingly, in case that a user inputs a command including a query for selecting a search result, it is necessary that a search server determines whether the command is inputted with malicious intent. In case that the command is inputted with malicious intent as a result of determination, information associated with the search result according to the query but the query may not be used as a standard to generate a search result list. As described above, it is require to determine whether “a command” including a query of a user, which is inputted to a predetermined system, is generated with malicious intent of some users, which is different from that the system aims at. A command inputted with malicious intent is preferable to be determined to be “invalid” and an original effect by the command (for example, adding the preference to the search result, as described above) may not be generated.
Korean Patent Application No. 10-2002-7010554 (Title of the invention: A System and method to determine the validity of an interaction on a network, hereinafter referred to as “cited reference 1”) discloses various method of determining whether a command is inputted with malicious intent, in addition to disclosed technology.
The cited reference 1 indicates an interaction inputted with malicious intent of a user as “an illegal interaction” in the specification.
The cited reference 1 includes 1) collecting data including “collective type data’ and “a unique characteristic data” is collected from a user interaction on network, 2) storing data in a database, 3) constructing an estimated model by the collective type data and the unique characteristic data in order to identify an illegal interaction with a network, 4) identifying an illegal interaction in the database by using the estimated model.
The cited reference discloses ‘a search listing click/the number of a user ID per unit time’, ‘entry source/the number of a user ID per unit time’, and ‘an advertiser accepting a click that is possible to request/the number of a user ID per time unit’ as “the collective type data”.
Also, the cited reference 1 discloses ‘the data of a click generating income’ and ‘a time stamp of a click generating income’ as “the unique characteristic data”.
Also, in case that an interaction is inputted, the cited reference 1 determines whether the interaction is valid by using the estimated model. Namely, “acceptable but uncommon class (ABUC) value”, “normal behavior class (NBC) value”, and “unacceptable class (UC) value” ┌with respect to an interaction┘ are respectively computed, and the interaction is determined to be a class having the largest value among them.
However, the cited reference 1 treats an interaction having “acceptable but uncommon class value” as a valid interaction, and there is a room for doubt that the interaction may be suspected to be an invalid interaction from the term of “acceptable but uncommon class value”.
Accordingly, according to the cited reference 1, after the interaction in a category of valid but still suspected to be invalid is determined to be “valid” once, the interaction becomes a valid interaction. Therefore, an illegal interaction is possible to be determined to be a valid interaction.
The cited reference 1 tries to solve the problem described above by periodically revaluating a mathematical equation to compute the ABUC value, the NBC value, and the UC value and generating a new model. However, since the solution is applied to only a new interaction newly inputted after an interaction that may be an illegal interaction is determined to be valid, a previous determination result can not be changed.
Accordingly, a new validity determination method is required, which can solve the problem of the prior art and can precisely determine whether a command is valid.